INDEX

Explanations

conditional statements or possibilities indicated by the word "could"

oai_token-act-pair · gpt-3.5-turbo

the word "could" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ing

-0.67

cedented

-0.60

 Rising

-0.59

 Forgotten

-0.58

 vacated

-0.57

velop

-0.57

kar

-0.56

 Cortex

-0.55

holder

-0.54

pite

-0.54

POSITIVE LOGITS

 feas

1.07

ĸļ

1.06

adian

1.03

berra

1.00

 tremend

0.88

 conce

0.87

 attest

0.86

 conclud

0.86

be

0.85

 theoretically

0.84

Activations Density 0.085%

conditional statements or possibilities indicated by the word "could"

the word "could" in various contexts

No Comments

No Known Activations